cabbage

joined 2 years ago
[–] cabbage@piefed.social 1 points 2 days ago* (last edited 2 days ago)

Large language models and "generative AI" such as Stable Diffusion, Midjourney, and DALL-E are all just machine learning models. We do not currently have a real "AI branch" of computer science, we have a branch of machine learning that poses as AI.

No matter how good a machine gets at recognizing and predicting patterns, it will not constitute AI, as intelligence is different from pattern recognition and prediction. Even if LLMs can sometimes appear to be reasoning, they importantly are not.

[–] cabbage@piefed.social 138 points 3 days ago (2 children)

five minutes later

Grok: "Heil hitler!"

[–] cabbage@piefed.social 11 points 4 days ago (1 children)

I guess this depends on things like location, soil type, and environmentally disruptive neighbours.

I grew up in a Scandinavian forest with a well, no filtering needed. Sure, after heavy rainfall the water could end up containing a few more minerals than usual, but it would never require filtering, and the house has been standing for almost 40 years now without sediments building up.

[–] cabbage@piefed.social 16 points 4 days ago

If they continue like this, their customers will be so fed up with them that they can lay off the entire customer-facing part of the company within a few years! Imagine how much money they can save once they don't have to deal with customers any more. Finally the AI innovation department will be able to focus fully on their work.

[–] cabbage@piefed.social 2 points 4 days ago

Yeah, I just find it to be a great rule of thumb. Those who understand what they are doing will be aware that they are not dealing with AI, those who jump to label it as such are usually bullshit artists.

[–] cabbage@piefed.social 4 points 4 days ago (4 children)

Usually when I see this it's using other machine learning approaches than LLM, and the researchers behind it are usually very careful not to use the term AI, as they are fully aware that this is not what they are doing.

There's huge potential in machine learning, but LLMs are very little more than bullshit generators, and generative AI is theft producing soulless garbage. LLMs are widely employed because they look impressive, but for anything that requires substance machine learning methods that have been around for years tend to perform better.

If you can identify cancer in x-rays using machine learning that's awesome, but that's very seperate from the AI hype machine that is currently running wild.

[–] cabbage@piefed.social 31 points 5 days ago* (last edited 5 days ago) (7 children)

Gigantic hater of all things LLM or "AI" here.

The only genuine contribution I can think of that LLMs have made to society is their translation capabilities. So even I can see how a fully open source model with "multilingual fluency in over 1,000 languages" could be potentially useful.

And even if it is all a scam, if this prevents people from sending money to China or the US as they are falling for the scam, I guess that's also a good thing.

Could I find something to hate about it? Oh yeah, most certainly! :)

[–] cabbage@piefed.social 2 points 5 days ago

In the sense of multiple users in Android settings? That works, it can be enabled in settings -> system -> multiple users. I haven't tested it though, as I don't have any need for that.

I use Microsoft Authenticator and Microsoft Outlook for work, and both work flawlessly with /e/OS. Thankfully I have not had any reason to test Teams, but I'm pretty sure that would work as well.

 

A call for help from @eighthave@librem.one, who among other things is associated with F-droid:

An open call to #Android #developers! The #EuropeanCommission needs help evaluating #GooglePlay's #security claims. I'm going to do what I can. Anyone with knowledge of how app installation, uninstallation, sandboxing, signing, etc. could really help here. If you want to contribute, please reach out!

I figured maybe someone around here could be capable of helping out. Note that his Mastodon instance, librem.one, does not support private messages.

[–] cabbage@piefed.social 4 points 6 days ago

That should suffice!

[–] cabbage@piefed.social 8 points 6 days ago* (last edited 6 days ago) (5 children)

I think making the rules more visible might also be on the todo list - they were just recently drafted after a discussion with the community. :)

[–] cabbage@piefed.social 22 points 6 days ago* (last edited 6 days ago) (7 children)

Great questions - I'm curious as well. Would also be useful to know as I'd be happy to volunteer as a mod or up my donations if necessary.

Regarding question 3, the answer is refreshingly clear. Here's from the rules:

PieFed is a place for thoughtful discussion, respectful debate, and factual engagement. We encourage diverse voices and aim to prevent content that harms, divides, or misleads. Accordingly the following types of content are prohibited on PieFed:

[...]

3 - Authoritarian and Fascist Propaganda

Content promoting fascist, ultranationalist, or authoritarian ideologies is not allowed, including things such as:

  • Genocide denial or minimization, such as denying or justifying atrocities in Gaza, Xinjiang, Ukraine, or elsewhere.
  • Use of fascist symbols & memes (e.g., Pepe the Frog), or 4chan/8chan screenshots.
  • Support for regimes or ideologies that suppress basic human rights including but not limited to Trumpism, Putin, Hamas, North Korea, Proud Boys.

There's also the Rational Discourse Toolkit in the side bar, which I think recognizes the role of social media as part of the problem and the challenge of trying to become part of the solution.

As for defederation, there's point 9 of the rules: "Generally this will be a last resort when problems from an instance are systemic or caused by the admins of the instance."

[–] cabbage@piefed.social 13 points 6 days ago (4 children)

I believe /e/OS supports a broader range of devices, and it's also pretty great in my experience. The focus is on getting rid of google (replacing all services with MicroG and nextcloud integration) and blocking trackers while providing a smooth user experience, so it's security features are not as over the top as Graphene. It's still a huge freaking improvement over stock Android though, and I find it to be a joy to use.

On devices supported by the online installer it can be up an running in like 30 minutes, no technical skills required. :)

 

The thing I hate the most about AI and it's ease of access; the slow, painful death of the hacker soul—brought not by war or scarcity, but by convenience. By buttons. By bots. [...]

There was once magic here. There was once madness.

Kids would stay up all night on IRC with bloodshot eyes, trying to render a cube in OpenGL without segfaulting their future. They cared. They would install Gentoo on a toaster just to see if it’d boot. They knew the smell of burnt voltage regulators and the exact line of assembly where Doom hit 10 FPS on their calculator. These were artists. They wrote code like jazz musicians—full of rage, precision, and divine chaos.

Now? We’re building a world where that curiosity gets lobotomized at the door. Some poor bastard—born to be great—is going to get told to "review this AI-generated patchset" for eight hours a day, until all that wonder calcifies into apathy. The terminal will become a spreadsheet. The debugger a coffin.

Unusually well-written piece on the threat AI poses to programming as an art form.

 

European Union watchdogs fined Apple and Meta hundreds of millions of euros Wednesday as they stepped up enforcement of the 27-nation bloc’s digital competition rules.

The European Commission imposed a 500 million euro ($571 million) fine on Apple for preventing app makers from pointing users to cheaper options outside its App Store.

The commission, which is the EU’s executive arm, also fined Meta Platforms 200 million euros because it forced Facebook and Instagram users to choose between seeing ads or paying to avoid them.

 

Swedish journalist Joakim Medin has been detained in Turkey after arriving in Istanbul on Thursday to report on the current protests.

Since Medin notified his editor in chief that he was being taken in for interrogation – around lunchtime on Thursday – nobody has heard from him.

Joakim Medin is a regular contributor to daily newspaper Dagens ETC and has reported from and about Turkey regularly for several years. We do not as yet know where or how he is, and we are imploring the Swedish government to take action to help us get Joakim home.

The chief editor of Dagens ETC wrote the following commentary on the decision to make this public, machine translated using FireFox:

This is what Dagens ETC's reporter Joakim Medin wrote yesterday when he landed in Istanbul to monitor recent developments:

"They are taking me in for interrogations now"

Now it's been more than 24 hours.

Silence.

I don't know where he is.

I don't know how he's doing.

Not if he's interrogated.

Not if he's going to be deported.

We at Dagens ETC have feverishly tried to get answers in different ways (the UD and consulate are connected).

But a whole day...

Now we choose to do this in public.

For Joakim to be released.

Because freedom of the press is under attack.

I demand that the Swedish government act directly and sharply.

I expect the support of just about every other media.

Joakim went to Turkey to give Dagens ETC's readers all the nuances.

Now he can't do that.

With the hope of very quick solution,

Andreas Gustavsson, Editor-in-chief Dagens ETC

 

I just saw this post (with video) from @DavidMadeThis@mastodon.social on Mastodon, and figured it might be worth sharing. Looks pretty neat.

Hello Mastodon! I'm a power engineer who is trying out solo game development and on the way to release Power Network Tycoon - a game where you build and manage your own electrical grid with real physics simulation.

If you've ever wondered how power systems actually work (or why they fail), this might be your jam. I built it to be technically accurate while still being fun. To say it's been a challenge is an understatement...

 

With two months left in office, the president for the first time authorized the Ukrainian military to use the system known as ATACMS to help defend its forces in the Kursk region of Russia.


President Biden has authorized the first use of U.S.-supplied long-range missiles by Ukraine for strikes inside Russia, U.S. officials said.

The weapons are likely to be initially employed against Russian and North Korean troops in defense of Ukrainian forces in the Kursk region of western Russia, the officials said.

Mr. Biden’s decision is a major change in U.S. policy. The choice has divided his advisers, and his shift comes two months before President-elect Donald J. Trump takes office, having vowed to limit further support for Ukraine.

Allowing the Ukrainians to use the long-range missiles, known as the Army Tactical Missile Systems, or ATACMS, came in response to Russia’s surprise decision to bring North Korean troops into the fight, officials said.

Mr. Biden began to ease restrictions on the use of U.S.-supplied weapons on Russian soil after Russia launched a cross-border assault in May in the direction of Kharkiv, Ukraine’s second-largest city.

To help the Ukrainians defend Kharkiv, Mr. Biden allowed them to use the High Mobility Artillery Rocket System, or HIMARS, which have a range of about 50 miles, against Russian forces directly across the border. But Mr. Biden did not allow the Ukrainians to use longer-range ATACMS, which have a range of about 190 miles, in defense of Kharkiv.

 

I remember when Trump first won, the American-centered part of the web I would occasionally stop by seemed completely infiltrated with MAGA trolls. You had this feeling people thought it was edgy and fun - the worst kind of cultural moment seemed to be happening.

This time around I'm not so much on mainstream social media. And when I do check them out, it seems hard to understand what the vibe is as most content is AI or from professional content creators.

The closest thing I see to Trump supporters these days seems to be the enablers who endlessly repeat how they won't vote for Harris for some dumb reason or another - they simply cannot vote for a black woman president because it's not progressive enough, and all that jazz. But I don't ever see Trump supporters.

Of course they exist still. I have just chose social media platforms strategically to avoid toxic people.

So I'm wondering if the same enthusiasm for Trump that seemed to be boiling online in 2016 is still there today, and if this election only feels different because I'm self-selected into saner platforms. Or if it is really different this time around.

I get that it's an incredibly difficult question to answer, but I would love perspectives from people who have kept up an active use of mainstream social media, or otherwise have some insights I lack.

 

Norway asks the International Court of Justice (ICJ) to clarify whether Israel is violating international law. On Monday, Israel adopted new laws that effectively ban the UN aid organization for Palestinian refugees.


Norway, the UN, the USA and several countries have reacted strongly to the fact that the majority in the Knesset passed two new laws on Monday evening which will ban the UN aid organization for Palestinian refugees, UNWRA.

One law prohibits UNWRA from operating in Israel and the Israeli-occupied territories from next year.

The second law stipulates that Israel labels the UN aid organization as a terrorist group and breaks diplomatic ties with the organization.

This means that the Israeli authorities are prohibited from having any contact whatsoever with the UN organisation.

Foreign Minister Espen Barth Eide (Ap) said in a press release yesterday that Norway strongly distanced itself from the decision.

Now the Norwegian government is going one step further.

Eide wants the UN's highest court to assess whether Israel is violating international law when they want to ban UNRWA's work.

- No one is above the law and no one is above international law. The occupying power Israel has a duty to facilitate support for the people living in Palestine under occupation. Therefore, we believe this decision is simply illegal, says Eide to NRK.

According to the government, a number of countries have expressed support for the initiative, in addition to the United Nations Organization for Palestine Refugees (UNRWA).

Doesn't think anyone can replace UNRWA now

Norway has long been critical of how Israel has operated in Gaza and the West Bank after the Hamas attack in October last year.

Following Israel's decision to ban the UN aid organisation, Norway fears that the consequences could be dramatic for hundreds of thousands of civilians.

The organization has provided aid to millions of displaced Palestinians for over 70 years. They have also continued to work, while the war has raged in Gaza.

Eide believes that the Israeli government is now making it difficult for the Palestinians to get vital help and basic services such as health care and school.

The Israeli authorities say they are working on other solutions that can replace UNRWA, but Barth Eide believes that is not good enough.

- All the important emergency aid organizations are clear that they need the UN's emergency aid organization for Palestinian refugees. They are the ones who have the experience, they are the ones who have the expertise and those who have the entire infrastructure both in Gaza and the West Bank, says Eide to NRK.

He does not believe that anyone can replace the UN organization in the current situation.

- I do not believe that there is an alternative plan for this that can be put in place in time. The need and the crisis are now, not in the future. So this must be reversed, says Eide.

- Undermines the work on a two-state solution

Eide also believes that Israel's decision could undermine the work for a viable Palestinian state and a two-state solution.

Israeli Prime Minister Benjamin Netanyahu, on the other hand, is well satisfied with Monday's decision.

- UNRWA staff involved in terrorist activities against Israel must be held accountable, he wrote on X after the new laws were passed.

Israel has claimed that 450 UNRWA staff worked for militant groups in Gaza and that several participated in the terrorist attack on Israel on 7 October last year.

Several countries put payments to UNRWA on hold. Norway was not among them.

An independent commission of inquiry later determined that Israel lacked evidence. UNRWA also fired around 20 staff for having a role in the 7 October attack.


Via NRK, the Norwegian public broadcaster.

2
submitted 1 year ago* (last edited 1 year ago) by cabbage@piefed.social to c/piefed_meta@piefed.social
 

I noticed responding to posts in communities hosted at lemmy.ml gives the following warning:

This post is hosted on lemmy.ml which will ban you for saying anything negative about China, Russia or Putin. Tread carefully.

While I see where this is coming from and I agree with the general sentiment, I'm not sure it's a great idea to include such a message. I basically read it as an invitation to be off-topic and to derail conversations in order to annoy the admins. While it comes from a point of good intentions, it can be disheartening for the people running communities on Lemmy.ml to receive comments about Russia from users basically trying to get banned, in communities that has nothing to do with this issue.

It's unfortunate, but a lot of valuable older communities are still hosted on lemmy.ml, and I think PieFed users should be encouraged to be constructive and on-topic users there as they should be everywhere else.

An alternative suggestion: Maybe it could be useful to remind people which community they are posting in? Like, "This community is dedicated to renewable energy. Please keep this in mind when contributing to the discussion". Then again, that would be a mess to implement in a good way.

1
submitted 1 year ago* (last edited 1 year ago) by cabbage@piefed.social to c/piefed_meta@piefed.social
 

Hi,

The CSAM scandal the other day got me thinking about the (often lacking) capability of the Threadiverse to deal with quickly with content moderation, and since PieFed has already been a bit experimental in this regard, I figured maybe this is a place where I could ask if an idea is feasible. Sorry if it's a bad match!

The idea is to identify trusted users, in the same way that PieFed currently identifies potentially problematic users. Long term users with significantly more upvotes than downvotes. These trusted users could get an additional option to report a post, beyond "Report to moderator": Something like "Mark as abuse".

The user would be informed that this is meant for content that clearly goes against the rules of the server, that any other type of issue should be reported to moderators, and that abuse of the function leads to revoke of privilege to use it and, if intentional, potentially a ban.

If the user accepts this and marks a post as abuse, every post by the OP of the marked post would be temporarily hidden on the instance and marked for review by a moderator. The moderator can then choose to either 1) ban the user posting abusive material, or 2) make the posts visible again, and remove the "trusted" flag of the reporting user and hence avoiding similar false positives in the future.

A problem I keep seeing on the threadiverse is that bad content tends to remain available too long, as many smaller instances means that the moderating team might simply all be asleep. So this seems like one possible way of mitigating that. Maybe it's not technically feasible, and maybe it's just not a particularly good idea; it might also not be a particularly original idea, I don't know. But I figured it might be worth discussing.

 

Congratulations on having made such a great tool, even in its early phase! It seems very solid.

I'm curious about the long-term plans for the project: Is the idea to work strictly with the Threadiverse (similar to Lemmy), or are there plans to integrate more with the microblog platforms (similar to Kbin)? Any particular difference in approach to Fediverse integration vis-a-vis the two main platforms?

view more: next ›